PiCSRL: Physics-Informed Contextual Spectral Reinforcement Learning

arXiv cs.LG / 3/31/2026

📰 NewsSignals & Early TrendsIdeas & Deep AnalysisModels & Research

共有:

Key Points

PiCSRL is proposed as a physics-informed contextual spectral reinforcement learning framework to enable more effective adaptive sensing from high-dimensional, low-sample-size (HDLSS) data where labeled labels are scarce.
The method injects domain knowledge by building embeddings and directly mapping physics-informed features into the RL state representation, alongside an uncertainty-aware belief model for improved prediction quality.
Evaluated on NASA PACE hyperspectral imagery of Lake Erie for cyanobacterial gene concentration sampling, PiCSRL reports substantially better station selection performance than random and UCB baselines.
Ablation results indicate that physics-informed features improve semi-supervised test generalization, with reported gains over using raw spectral bands alone.
Scalability experiments suggest PiCSRL can handle large network settings (e.g., 50 stations and over 2M combinations) while still outperforming baselines with statistical significance.

Abstract

High-dimensional low-sample-size (HDLSS) datasets constrain reliable environmental model development, where labeled data remain sparse. Reinforcement learning (RL)-based adaptive sensing methods can learn optimal sampling policies, yet their application is severely limited in HDLSS contexts. In this work, we present PiCSRL (Physics-Informed Contextual Spectral Reinforcement Learning), where embeddings are designed using domain knowledge and parsed directly into the RL state representation for improved adaptive sensing. We developed an uncertainty-aware belief model that encodes physics-informed features to improve prediction. As a representative example, we evaluated our approach for cyanobacterial gene concentration adaptive sampling task using NASA PACE hyperspectral imagery over Lake Erie. PiCSRL achieves optimal station selection (RMSE = 0.153, 98.4% bloom detection rate, outperforming random (0.296) and UCB (0.178) RMSE baselines, respectively. Our ablation experiments demonstrate that physics-informed features improve test generalization (0.52 R^2, +0.11 over raw bands) in semi-supervised learning. In addition, our scalability test shows that PiCSRL scales effectively to large networks (50 stations, >2M combinations) with significant improvements over baselines (p = 0.002). We posit PiCSRL as a sample-efficient adaptive sensing method across Earth observation domains for improved observation-to-target mapping.